18.6K ·GitHub
What is this?
The Llama Cookbook — Meta's official collection of recipes for building with the Llama model family: inference, fine-tuning, RAG and end-to-end use cases, mostly written as Jupyter notebooks.
The Llama Cookbook — Meta's official collection of recipes for building with the Llama model family: inference, fine-tuning, RAG and end-to-end use cases, mostly written as Jupyter notebooks.Is it a library I install, or a collection of example notebooks?A recipe collection rather than a package: notebooks organized by topic, plus a src folder holding the original llama-recipes library code and fine-tuning FAQs.
A recipe collection rather than a package: notebooks organized by topic, plus a src folder holding the original llama-recipes library code and fine-tuning FAQs.How do I actually run the recipes?There is nothing to install — you open the notebooks and run them. Start with the Getting Started section, the reference for inferencing, fine-tuning and RAG, then explore the provider-integration and end-to-end use-case folders.
There is nothing to install — you open the notebooks and run them. Start with the Getting Started section, the reference for inferencing, fine-tuning and RAG, then explore the provider-integration and end-to-end use-case folders.What do I need on my machine before opening a notebook?Python with Jupyter Notebook. Each recipe adds its own dependencies — the repo covers stacks like LangChain and vLLM — so check the requirements of the specific notebook you open.
Python with Jupyter Notebook. Each recipe adds its own dependencies — the repo covers stacks like LangChain and vLLM — so check the requirements of the specific notebook you open.Anything flaky about the repo right now?The repo was recently refactored, so some links are broken and a few folders are missing — the FAQ acknowledges this and points to the archive-main branch for the pre-refactor layout.
The repo was recently refactored, so some links are broken and a few folders are missing — the FAQ acknowledges this and points to the archive-main branch for the pre-refactor layout.Is the code free to use in my own projects?The repo itself is MIT-licensed per GitHub metadata, but the models are not: every Llama generation from 2 to 4 ships its own Meta license file plus an Acceptable Use Policy you must accept to use the models.
The repo itself is MIT-licensed per GitHub metadata, but the models are not: every Llama generation from 2 to 4 ships its own Meta license file plus an Acceptable Use Policy you must accept to use the models.Which Llama models do the recipes cover?Llama Text and Vision models. The newest recipes target Llama 4 — Scout and Maverick — and the repo keeps license files for every generation back to Llama 2.
Llama Text and Vision models. The newest recipes target Llama 4 — Scout and Maverick — and the repo keeps license files for every generation back to Llama 2.What Llama 4 recipes can I try today?Getting started with the Llama API, a WhatsApp integration through that API, 5M-token long context with Llama 4 Scout, and research-paper analysis or character mind maps with Llama 4 Maverick.
Getting started with the Llama API, a WhatsApp integration through that API, 5M-token long context with Llama 4 Scout, and research-paper analysis or character mind maps with Llama 4 Maverick.Is this the same project that used to be called llama-recipes?Yes — Meta renamed llama-recipes to llama-cookbook and refactored the repository; the pre-refactor snapshot is preserved on the archive-main branch.
Yes — Meta renamed llama-recipes to llama-cookbook and refactored the repository; the pre-refactor snapshot is preserved on the archive-main branch.What is the difference between the Getting Started and End to End sections?Getting Started is the reference: inference, fine-tuning and RAG examples for learning the core skills. End to End Use Cases are complete applications across domains, while 3P Integrations collects recipes from various Llama providers.
Getting Started is the reference: inference, fine-tuning and RAG examples for learning the core skills. End to End Use Cases are complete applications across domains, while 3P Integrations collects recipes from various Llama providers.What are similar projects?claude-cookbooks, alpaca-lora, ai-engineer-notebooks, unsloth, MiniCPM, self-llm, Llama-Chinese, OpenLLM
claude-cookbooks, alpaca-lora, ai-engineer-notebooks, unsloth, MiniCPM, self-llm, Llama-Chinese, OpenLLMWhich topics is it in?finetuning, llama2, long-context, vision, whatsapp, jupyter-notebook, llama, examples, langchain, vllm, jupyter notebook, fine-tuning
finetuning, llama2, long-context, vision, whatsapp, jupyter-notebook, llama, examples, langchain, vllm, jupyter notebook, fine-tuning